Total Allowable Error: Important Concepts for Clinical Laboratories and a List of Recommendations
Written by Data Visualizations on . Posted in Uncategorized
It’s possible that you’ve reviewed method performance information that includes an arbitrary acceptance criterion of plus or minus 10%. While that might seem fine, it’s actually rather unfortunate because different measurands vary in the magnitude of error than affects clinical decision making. For example, troponin needs more accurate and careful measuring than a substance like lactate dehydrogenase (LDH).
Westgard and associates developed the practice of total analytical error (TAE) all the way back on 1974. The purpose of developing this practice was to improve upon the practice of judging method performance that was in place at the time. They noticed that it was almost always a better practice to judge total error instead of judging factors like accuracy and precision as separate categories. So, for instance, a method that is incredibly precise can tolerate a greater margin of inaccuracy and vice versa. When all of the errors associated with a single assay are added up, we’re left with total analytical error. A concept related to this directly is total allowable error (ATE). This error is the maximum analytical error we can tolerate and still determine clinically relevant differences between results. In layman’s terms, that means the test is still clinically useful.
Now, it’s time to talk about total allowable error as a means of maintaining good assay performance. In other words, we’re talking about total allowable error as an “error budget.” First and foremost, we need to define total allowable error for each assay. The laboratory director might be the best person to determine what these limits are. Even better, rely on CLIA regulations, external proficiency testing programs, performance goals based on biological variation, and/or goals from professional organizations and expert committees. Once that’s all set, it’s possible to allocate allowable error to different sources of error. It is important to take note of potential biases while you’re doing this, though! Remember: we need to leave the majority of the total allowable error budget up for random variables.
Finding the right total allowable error recommendations can be tough. That’s why we created this short guide for you to gain a better understanding. We hope this information was useful to you.